Interactive Visualization using Linked Views

purpose is to highlight an approach to spatial expression analysis using ideas from interactive visualization
high-level idea: both interactive and direct visualization give us something like maps. We want to be able to inspect them simultaneously.
We want to generate these interactive views in demand, so we wrap the core visualization code in an htmlwidgets package (https://www.htmlwidgets.org/).
First, we"ll install some packages needed to prepare the data and run the analysis.
We use spatial data analysis packages to prepare the geojson which ends up as input to our visualization

library("dplyr")

## 
## Attaching package: 'dplyr'

## The following objects are masked from 'package:stats':
## 
##     filter, lag

## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union

library("geojsonio")

## 
## Attaching package: 'geojsonio'

## The following object is masked from 'package:base':
## 
##     pretty

library("raster")

## Loading required package: sp

## 
## Attaching package: 'raster'

## The following object is masked from 'package:dplyr':
## 
##     select

library("readr")
library("rmapshaper")

## Registered S3 method overwritten by 'geojsonlint':
##   method         from 
##   print.location dplyr

library("umap")
library("BIRSBIO2020.scProteomics.embeddings")

Next, we load the MIBI-ToF data or download it, if it’s not available.

data_dir <- file.path(Sys.getenv("HOME"), "Data")
download_data(data_dir)

## Warning in dir.create(directory, recursive = TRUE): '/github/home/Data' already
## exists

## [[1]]
## [1] "/github/home/Data/mibiSCE.rda"
## 
## [[2]]
## [1] "/github/home/Data/masstagSCE.rda"
## 
## [[3]]
## [1] "/github/home/Data/TNBC_shareCellData"

loaded_ <- load_mibi(data_dir)
cell_full <- read_csv(file.path(data_dir,  "TNBC_shareCellData", "cellData.csv"))

## Parsed with column specification:
## cols(
##   .default = col_double()
## )

## See spec(...) for full column specifications.

Our visualizations are going to be made on a patient by patient basis.
In principle, a similar approach should work while looking at many patients simultaneously
- We can still make a u-map on all patients
- But we’d need to layout the spatial data so that many patients can be displayed next to one another.
- Would also be better to group the panels for patients, so that patients with more similar prognosis are placed nearer to one another
Since we’ll focus on one patient at a time, we’ll want to re-use the data preparation code
This is encapsulated in the function below. The input are the full protein expression data, paths to all the TIFFs, and the desired patient to display.
The output are a dataframe giving the coordinates of each cell, according to a u-map, and a geojson of the cells
- We simplify the geojson to make interaction smoother.
- We also crop the rasters, so that preprocessing is faster (and so we can display more patients in an rmarkdown without requiring too much time to compile)

input_data <- function(cell_full, tiff_paths, sample_id, crop_size=1000, simplify=0.4) {
  # read the associated raster
  im <- raster(tiff_paths[grepl(paste0("p", sample_id, "_"), tiff_paths)])
  im <- crop(im, extent(im, 0, crop_size, 0, crop_size)) # comment out to work on full data
  polys <- polygonize(im) %>%
    filter(!(cellLabelInImage %in% c(0, 1)))

  # filter down to relevant data
  cell_data <- cell_full %>%
    filter(
      SampleID == sample_id,
      cellLabelInImage %in% polys$cellLabelInImage
    )

  polys <- polys %>%
    left_join(cell_data)

  # run U-Map on channel data
  channels <- setdiff(colnames(cell_data), c("cellLabelInImage", "SampleID", "cellSize", "Background", "tumorYN", "cell_cluster", "tumorCluster", "Group", "immuneCluster", "immuneGroup"))
  dimred <- umap(cell_data[, channels]) %>%
    .[["layout"]] %>%
    data.frame() %>%
    rename(V1 = X1, V2 = X2)
  dimred <- cbind(cell_data, dimred)

  # convert to geojson
  geo_polys <- geojson_json(polys) %>%
    geojson_sp() %>%
    ms_simplify(keep = simplify) %>%
    geojson_json()

  list("dimred" = dimred, "polys" = geo_polys)
}

Below are the results.
Each side by side panel is a subset of cells from one region of the sample from a single patient.
How to read the tool: left hand side are the cells, as they appear in the MIBI ToF data. They are shaded in by their cell type.
- htmlThe tumor clusters (from the original rda object) are 4, 7, 10, 17.
- htmlThe immune clusters are 1, 2, 3, 4, 8, 10, 11, 12.
How to use the tool: Draw a brush over one panel to highlight the corresponding cells in the other

data_ <- list()
for (i in seq_len(6)) {
  data_[[i]] <- input_data(cell_full, loaded_$tiffs, i)
}

linked_views(data_[[1]]$polys, data_[[1]]$dimred)

linked_views(data_[[2]]$polys, data_[[2]]$dimred)

linked_views(data_[[3]]$polys, data_[[3]]$dimred)

linked_views(data_[[4]]$polys, data_[[4]]$dimred)

linked_views(data_[[5]]$polys, data_[[5]]$dimred)

linked_views(data_[[6]]$polys, data_[[6]]$dimred)

ksankaran@wisc.edu ↩︎

Kris Sankaran1

Kris Sankaran¹